The problem
Large general-purpose AI models demand substantial computation and energy. Domain-specific applications motivate smaller systems that can reason, learn, and make decisions under practical resource constraints.
What we did
We outlined pathways toward energy-efficient, domain-specific AI models and agents, connecting algorithm design, brain-inspired computing, and hardware–software co-design. The work examines how these layers can jointly support more capable and efficient systems.
Why it matters
This is a research perspective on future AI systems. Its efficiency ambitions describe directions for model and hardware development rather than a measured speedup from a single implementation.
Paper
Abhijit Chatterjee, Niraj Jha, Jonathan D. Cohen, Thomas L. Griffiths, Hongjing Lu, Diana Marculescu, Ashiqur Rasul, Keshab K. Parhi. “Energy-Efficient Domain Specific Artificial Intelligence Models and Agents: Pathways and Paradigms.” Listed in the resume as ACM Computing Surveys, 2025.